W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > January 2002

RE: TDL Model Theory

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Date: Thu, 24 Jan 2002 14:50:59 -0000
To: "Graham Klyne" <Graham.Klyne@MIMEsweeper.com>
Cc: <w3c-rdfcore-wg@w3.org>
Message-ID: <JAEBJCLMIFLKLOJGMELDEENHCCAA.jjc@hplb.hpl.hp.com>
> I have a couple of comments/questions...

I'll try and answer.

> OK, they key idea here is that the denotation of a literal node is a
> <unicode-string,value> pair, rather than just a value, right?


> So the denotation of a node labeled with "10" and type-designated to be a
> decimal integer can never be exactly the same as a node the
> denotes just an
> integer value that is, say, the count of my fingers-and-thumbs, without
> regard for how it is lexically represented.   (I don't see this as a
> problem, but note that it seems like a departure from previous
> approaches.)

Yes, but such nodes don't exist in the models ... unlike S-A, (or the
original P), at least in the model theory all values (such as integers) are
always paired with a lexical representation.

> >The Interpretation of Datatype URIs
> >
> >If E is a uriref and the label of E=u(d) for some d in DT, then I(E) = d.
> Hmmm... it's just a nit (I think) but this implies that a uriref has a
> label.  Do you mean something like:  "If E is labelled with a uriref and
> the label of E=u(d) ..."

Yes that's fine I accept the nit.

> >The Interpretation of Blank Nodes
> >The mapping A on blank nodes is unrestricted and a blank node can be
> >interpreted
> >as any object in the Universe (including literal-value pairs).
> >
> >The Interpretation of Asserted Triples
> >
> >The function IEXT is modified as follows:
> >IEXT maps the set of properties IP into the powerset of ( Universe x
> >Universe).
> If literals cannot be subjects, would this become a powerset of (IR x
> Universe)?

I want rdf:value as the identity which is why its Universe x Universe.

I am not suggesting literals as subjects in any serialization of RDF, but in
terms to the graph, and particularly taking the schema closure of a graph
literal as subjects is necessary in this approach. (Both the P & D idiom
really does seem to have literals as subject or type statements, just not

> Something thing that is bothering me about this is:  the
> interpretation of
> a Unicode node is stated in such a way that there may be several
> literal-value pairs that could be denoted.  Do you mean the
> interpretation
> in this case to be ambiguous?

Extremely. I am trying to delete the suggestion that xsd:string is the
default type.

If we support no types, or no type information is available for a particular
node labelled "foo", then that node can be interpreted as a pair <"foo", x >
for any x what so ever.

It's model theory not application semantics.

 Suppose that data types BinaryNumber and
> DecimalNumber are recognized, then a node labelled "10" can denote:
>     <BinaryNumber,2>
>     <DecimalNumber,10>

"10" without type information could also just be the string, or maybe
there's its of some type we haven't thought of yet. So in the pair < "10", x
> an application is only really going to find the "10" helpful in this case,
since the x is free.

> Also, how does the change to rdf:type work for data types that
> don't have a
> defined lexical form?  E.g. consider the format used by RDFWeb for
> describing people:
>    _:gk rdf:type  foaf:Person .
>    _:gk foaf:name "Graham Klyne" .
>    _:gk foaf:mbox <mailto:GK@ninebynine.org> .
>    (etc...)
> There is no defined lexical form that universally denotes me, the
> person.  So what is the denotation of the thing labelled _:gk ?

Pass for now.


Received on Thursday, 24 January 2002 09:51:01 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:53:54 UTC